Hi Brad,
On Friday, 2007-03-30 17:41:43 +1000, Brad Hards wrote:
> > > The equality operator has a couple of unexpected (to me) test cases:
> > > =ISNA(NA()=NA()) is required to be true.
> > > I don't agree with the logic "If there's an error on either side, the
> > > result is an error -- even if you're comparing the "same" error on both
> > > sides.". That isn't intuitive, and makes for messy logic.
> >
> > Indeed, the description is correct, but the result value is not, it
> > should be Error instead. The NA() function generates an error and the
> > error is propagated through the expression, which is the usual behavior.
> That means that ISNA() can never return true - it can only ever be false (the
> argument was not an NA() error) or an Error (if the argument is an error).
> I'm not sure that makes sense. This is an intentional error check function.
Ah, so yes, I was mislead, =ISNA(NA()=NA()) must be TRUE because
=NA()=NA() is Error, that's what the comment refers to. So what in that
test case's comment do you actually disagree with?
Eike
